Rice Leaf Disease Identification Using Adam Optimizer Based Modified Differential Evolution Algorithm

Recently, one of the grain-based crops that is farmed the most is rice, which is important to the agricultural sector. Furthermore, rice is a significant crop that is consumed by almost half of the global population. However, biotic and abiotic elements including bacteria, viruses, pests, soil fertility, and so forth have an impact on agricultural production. Early disease identification saves farmers time, increases crop yield, and guards against production loss in rice plants. The rice fields used for cultivation are the source of the rice leaf image dataset. The three disease classes that made up the data acquisition were Brown Spot, Bacterial Blight, and Leaf Blast. The ultimate goal of this proposed study is to improve rice leaf detection and classification performance. Thus, using a deep learning approach, this study presented an Adam Optimizer (AO) to carry out an efficient classification of rice leaf disease. Moreover, rice leaf disease detection and classification involve the use of Transfer Learning in conjunction with Inception V3 architecture. The Modified Differential Evolution Algorithm (MDEA) technique is used to determine the transfer learning approach's optimal learning rate. Finally, the proposed AO-MDEA shows better performance metrics in terms of accuracy (98.74%), precision (99.04%), recall (99.12%) and F1-score (99.15%) respectively.